When different telecommunications network carriers exchange voice-over-IP traffic—for example, when a Voice-over-IP telephone call is made from a subscriber of a first carrier to a subscriber of a second carrier—the exchange of data is, in accordance with current practice, invariably performed with use of traditional Time Division Multiplexed (TDM) links. Meanwhile, the transmission of Internet Protocol (IP) traffic (i.e., network packets) within a given carrier is commonly performed with use of a packet loss concealment technique which recognizes, and compensates for, the loss of packets (i.e., the failure to receive one or more of the transmitted packets). However, such packet loss concealment techniques are far from perfect, and often introduce audible distortions in the resultant speech.
In addition, it is often necessary for network carriers to guarantee (or at least to be able to measure) a Quality-of-Service (QoS) level to (or for) its customers. In order to be able to do so when VoIP calls have been received from another carrier, it would be highly advantageous for the receiving carrier to be able to identify (e.g., count) the presence of packet losses which occurred in the other carrier's IP network, particularly those that have introduced such audible distortions. However, while Real-time Protocol (RTP) header information is used within an IP packet network to detect lost packets on IP networks, there are currently no methods for detecting whether such packet losses have occurred on speech that is no longer packetized.
Therefore, it would be highly desirable to be able to estimate a packet loss rate and pattern from a speech signal that has been encoded, transmitted through an IP network, decoded with the use of concealed packet loss techniques, and subsequently converted to a non-packetized form (e.g., TDM). In other words, it would be desirable to be able to determine packet loss that has occurred once the speech has been reconstructed and, therefore, lost packet information is no longer available.